to Confirm Göran Brusewitz
This study is an attempt to conceptually replicate a study by Ramakers, Stevens and Morris (2005) using a measure of electrodermal activity skin conductance (EDA) to evaluate the possibility of telepathy occurring between biologically and/or emotionally related senders and receivers. Ten negatively valenced and highly arousing target pictures were mixed with 10 blank control pictures in 10 blocks, with one of each kind in each block.
The order of presentation of the target and control pictures within the block was determined randomly by a computer program. The series of 20 pictures were shown for the sender on a computer screen. Relaxation for the receiver was facilitated by soft music. It was hypothesized that there would be significantly more variance in the receiver EDA when the sender was exposed to negative arousing pictures, than to blank pictures. The results failed to show a significant difference in EDA variance between negative arousing and blank pictures, and did thus not support the telepathy hypothesis. It was recommended that future replications allocate more time for relaxation for the receiver.
The goal of this study was to conceptually replicate the results previously reported by Ramakers et al. (2005) on the effect of remote emotion on receiver skin conductance.
The purpose of those experiments was to use a physiological measure to evaluate the possible occurrence of a telepathic transfer of emotions between senders and receivers.
Current findings and the present state of research in parapsychology with EDA are reviewed.
Parapsychology
Reports of psychic phenomena dealing with dreams coming true, premonitions, telepathy and healing are legend throughout human history. As they are reported, many of these experiences have been very dramatic and filled with emotions for the persons involved and often concerned vital issues such as death, injury and health. Naturally, it is just this universal nature of these themes together with the controversial nature of their interpretation which enables today‟s mass media to exploit these experiences in the form of films and pseudo-scientific documentaries. Beside this, many individual claimants may have a genuine wish to use their alleged ability to “help” others.
Nevertheless, despite the claims it often remains impossible to determine whether the effects they can create with this alleged ability are due to normal processes. For example, in the case of healing, can the effects an alleged healer can create, be attributed to spontaneous remission and placebo effects, or are they genuine and a result of a genuine ability?
---
I would like to thank John Björkhem Memorial Foundation for the financial support which made these experiments possible.
Whatever the nature of the effect is, it can lead to the claimant being convinced of possessing a genuine ability, a conviction that finds further support in confirmation bias and selective memory. Ultimately it is the task of research to resolve the issue whether the effect that seems paranormal is genuine or not, an issue which is undoubtedly an important one given the implications that hitherto unknown abilities might have for our world view. One very important step for the research was to introduce some basic concepts for the different phenomena that are reported by different individuals and studied by scientists. Telepathy is already mentioned, which stands for direct transference of information from one living organism to another, a mind-to-mind communication (Parapsychological Association FAQ File 1, 2008). Clairvoyance or remote viewing is the perception of events at remote locations, beyond the reach of the normal senses. Still another phenomenon is precognition, or premonition, standing for perception of future events not foreseeable by any means of known derivation, to get information about/from the future. All these phenomena are labelled extra-sensory perception, ESP, perception without the use of any of the known senses. The other kind of phenomenon in parapsychology is psychokinesis, PK, or today mentioned as mind- matter interaction, standing for direct mental interaction with physical objects, animate or inanimate. A special kind of PK is bio-PK when the object to be influenced is a biological, living organism. The last concept to introduce is psi, one that covers both ESP and PK, the anomalous process of information or energy transference (Bem &
Honorton, 1994). Psi is often used when it is difficult or impossible to know if ESP or PK may have been used.
In order to go further, methodologies are needed to determine whether or not an acclaimed healer or shaman or for that matter any given individual claimant has any real talent. In particular, the growth of public interest in healing and alternative and complementary medicine justifies the current efforts directed at developing research strategies which will enable us to evaluate the claims made for paranormal healing. It should be said that many successful experiments in parapsychology appear at least at a prima facie level to have been very carefully designed (Hyman, 1995; Irwin & Watt, 2007; Utts, 1995). However a bedevilling issue in this field is that of the so-called experimenter effect in which some researchers consistently obtain significant results while others equally consistently fail to do so. Some progress has been made towards resolving this issue by setting up joint experiments run by a sceptic and a proponent (Wiseman & Schlitz, 1997; Wiseman & Schlitz, 1999), to which we will return later.
Previous research and findings
There have been various research enterprises on these phenomena beginning from the early 1900s, but it is generally accepted that a major step forward was taken in the early 1930‟s by J B Rhine at Duke University, where systematic research was started on telepathy with so called Zener cards. The card guessing strategy used established statistical methods to evaluate if a participant using the cards in a telepathy experiment scored significantly better than chance (Rhine, 1938). Although this basic design has been retained, aspects of it have been continuously improved. (By replacing card guessing with targets such as film clips that are closer to other human experience, ecological validity has thought to have been improved. Other measures have been brought in to make the design more methodologically stringent).
The design which is most used today in psi experimentation is the ganzfeld method. The method as applied to psi research was developed first in the 70‟s by three researchers independent of each other – Honorton, Braud and Parker (Braud, Wood, & Braud, 1975;
Honorton & Harper, 1974; Parker, 1975). The idea behind this method is to reduce the perceptual input to the usual senses, thereby making the receiver more open to inner experience including possible telepathic ones (Parker, 1975). In order to induce this
“ganzfeld state”, a homogenous field of vision is created by translucent spheres placed over the eyes of the receiver and homogenous source of sound stimulation is also provided (for instance by sea shore noice). The aim is to facilitate in the receiver a relaxed psi-conducive state of consciousness with inwardly directed attention.
The receiver is required to report any resulting impressions which form a so-called mentation report. This is recorded and written down, and the receiver will afterwards compare these impressions with the film sequence that was chosen along with the three other film sequences belonging to the same set of films that had been randomly selected by a computer. There are in fact two randomisations: first, the set of film clips is randomly chosen, then one of the four series of film clips is selected as the target and the others become decoys. The receiver will on the basis of the ganzfeld experiences rank these films in the order number 1 – 4, with 1 for the film most corresponding to the impressions, 2 for the next sequence to correspond etc. By convention, the rank number 1 is regarded as a hit.
One major change occurred when Berger and Honorton (1985; Honorton et al., 1990) developed an automated ganzfeld testing system to eliminate potential methodological problems identified in earlier ganzfeld studies (Honorton, 1979; Hyman & Honorton, 1986; Kennedy, 1979). Until then, a manual ganzfeld system had been used:
“The autoganzfeld system was based on a computer controlled video cassette recorder (VCR) with a custom-built video switch that allowed the computer to control the VCR‟s video output to each of three TV monitors. In the autoganzfeld system there was automated target randomisation and record keeping as well as a highly automated procedure which had to be adhered to” (Goulding, Westerlund, Parker, & Wackermann, 2004, p 67).
In order to meet criticism about various shortcomings in the procedure, this method was improved and developed into a digitalized autoganzfeld system. One was developed at the University of Edinburgh and was also used at the University in Northampton (Goulding, Westerlund, Parker, & Wackermann, 2001; Westerlund, Parker, Dalkvist, &
Goulding, 2004), another one at Liverpool Hope University College. A digital autoganzfeld system basically relies on a computer, autoganzfeld software, and computer monitors. The most significant improvement which marks the autoganzfeld is that the whole procedure is now automatized with all the steps registered and this makes all the study more safeguarded against bias and cheating. In earlier ganzfeld experiments, the source for randomness could vary between random tables and Random Number Generator in the computer (Parker, personal communication, 2007-08-23). At the University of Gothenburg another model was developed, with the main difference that it allowed a real-time recording of the mentation report to be used during the judging procedure (Goulding et al., 2004).
Using this technology a file with the mentation from the receiver can also be sent after the experiment to an independent judge, to attempt to identify the target film clip seen by the sender. The judge will thus watch all four film clips and rate them with a scale from 0 (no correspondence between film content and mentation) to 100 (strong correspondence) (Goulding et al., 2004). The film clip with the lowest rating will thus be ranked as number 4, and the highest rating will be ranked as number 1. The task is aided by the synchronized recording of the mentation with the film clip (Mann, 2004;
Westerlund et al., 2004).
Meta analysis of experiments with ganzfeld
During the years of 1974-1999 more than 100 ganzfeld experiments have been carried out (Palmer, 2003). With the accumulation of all the later experiments, it became necessary to find a way to evaluate this research. Meta-analysis is a method, which endeavours to be an objective, statistical method to evaluate replicability in the majority of or all experiments in a special field of research. The method is not quite objective, since many subjective elements are part of the process: “For example, when is a study similar enough, conceptually or methodologically, to what might be called the population norm to be included in the sample” (Palmer, 2003, p 52), i.e. when is a study similar enough to the basic design, when does it belong to this special field of research?
When you are confronted with borderline examples, there seems to be no consensus how to judge them – are they to be included or not? John Palmer reminds us about another difficult problem involved in carrying out a meta-analysis:
“What outcome is necessary or sufficient to support the claim of replicability? Is a significant Stouffer Z enough? (....the p-value is converted to a z-score, the z-score are summed, the sum is divided by the square root of the number of studies, giving Stouffer Z...). Is a significant Stouffer Z unnecessary if the mean effect size from the second sample falls within the 95 % confidence intervals of the mean effect size of the first sample” (Palmer, 2003, p. 52).
This is an illustration Palmer uses to illustrate a replicability issue in meta-analysis with a 2-experiment sample. There have been cases where researchers have claimed a successful replication if the mean of the second experiment falls within the confidence interval of the first experiment (personal communication, Palmer, April 21, 2008)? Does the occurrence of a significant heterogeneity mean that replicability has failed irrespective of the other outcomes? The question if the ganzfeld-studies are replicable depends partly on how these questions are answered.
The first meta-analysis of ganzfeld experiments was done by Hyman (1985) and included 42 studies. He found that the result was statistically significant, but maintained that it could be due to methodological weaknesses. One critical point for Hyman was the so-called “file-drawer” problem (Palmer, 2003), a point well-known to meta- analysts. The idea is that unsuccessful studies tend not to get published, meaning that the published studies represent a positively biased sample of all those conducted. This possibility, that there are unsuccessful studies that have been performed but not published and therefore were not included in the meta-analysis, is rebutted by Palmer (2003) in three ways: 1) Blackmore (1980) made a survey of ganzfeld researchers and failed to find evidence for unpublished ganzfeld studies. Palmer also mentions 2) the policy of the Parapsychological Association that its affiliated journals not discriminate
against publication of negative results. The last way (and according to Honorton the most effective rebuttal) was 3) to employ a statistical technique developed by meta- analyst Robert Rosenthal (1979) to estimate the number of studies that would need to be in the file-drawer for the overall result of DH (direct-hit, studies where the receiver has four possible targets, ranking the right film clip target as number 1) studies to be reduced to non-significance:
“The number he (Honorton/GB) came up with was 423, or more than 12.000 sessions. Although 12.000 is probably an overestimate because Hyman‟s aborted studies would have low number of sessions, the idea that even half that number of sessions were conducted and not reported seems unlikely, particularly since a typical ganzfeld experiment takes about two hours to conduct and the number of parapsychologists is quite small” (Palmer, 2003, p. 56).
Hymans report was followed by a meta-analysis by Honorton (1985). He found 28 studies where direct hits were used as the major measure amongst these 42 studies. The analysis gave a highly statistical significant result (p = 10-11). Honorton noted that some studies had potential methodological weaknesses, but concluded that the significance did not depend on these weaknesses; even those with proper methodology gave significance. Following this was a joint report by Hyman and Honorton (1986) where they agreed upon methodological clear and stringent guidelines for future ganzfeld studies to follow. They both emphasized the importance of standardizing the procedure (Hyman & Honorton, 1986), but did not agree on how results from existing experiments should be interpreted.
Eight years later, Bem and Honorton (1994) published a meta-analysis on 11 autoganzfeld experiments that had been carried out using the guidelines from Hyman and Honorton. These experiments were partly automatized (Bem & Honorton, 1994) and the method used video clips as target stimuli, not only pictures. The result from this study gave significant support for the psi hypothesis with a mean hit rate on 32% (with 106 hits in the 329 sessions, z=2.89, p=0.002, one-tailed, which corresponds to an effect size of .59). This mean hit rate can be compared to the fact that chance mean expectancy would be 25% and is a little lower than earlier, but well in accordance with the earlier analysis by Honorton (1985). Hyman (1994, as mentioned by Palmer (2003)) did however criticize this meta-analysis on some points, one of them being it was based on experiments from a single laboratory.
A few years later Milton and Wiseman (1999) carried out a new meta-analysis, this one containing 30 ganzfeld studies from seven different laboratories which had published since the guidelines by Hyman and Honorton. This meta-analysis did not show any significant result. The mean hit rate was 27 %. Bem, Palmer and Broughton (2001) added to this analysis with ten new ganzfeld studies and reached with the exact significance level a significant result (a significant Stouffer z of 2.59, p=0.0048, with an effect size of 0.051) (Palmer, 2003) with a mean hit rate on 30%. The large difference between the non-significant meta-analysis by Milton and Wiseman (1999) and the significant by Bem et al. (2001) was that Bem et al. (2001) did include a very successful study by Kathy Dalton (1997). Milton had excluded this study, regarding it as an outlier, with an extreme result, that unduly made the distribution in the result uneven. Palmer disagreed very strongly with the notion here of outliers, even if this is practiced and justified or at least tolerated by certain meta-analysts: “My primary reason for this is
that removing outliers misrepresents the data, which is the cardinal sin in any data analysis”(Palmer, 2003, p 61). Certainly it is the case, as Palmer stated, that Dalton‟s study did contribute to a statistical finding that deserves to be taken seriously, namely that of heterogeneity. It was found that all three groupings of studies (carried through by Honorton in the 1980s at Psychophysical Research Laboratory (PRL), those before (pre- PRL) and those afterwards (post-PRL) are significantly heterogeneous (Palmer &
Broughton, 2000). For example, the p-values for heterogeneity for both studies with direct hits in pre-PRL-data and the 28 with direct hits in post-PRL-data are in fact 10-5 (Palmer, 2003). There is however a weakness that the significance in the later (well controlled) autoganzfeld studies does depend on one single researcher, namely Kathy Dalton. With her study there is significance, and without her study, there is not.
To return to the heterogeneity issue, Bem et al. (2001) maintained that some of the studies Milton and Wiseman (1999) used in their meta-analysis were far too heterogeneous. The designs varied too much, and in their analysis they mixed studies with music as stimuli, but also pictures or film sequences. The various meta-analysis become good examples of the difficulty in replicating previous meta-analysis, because if you have different selection criteria for the studies to be included, different results can emerge. Bem was interested in finding out how important it was to follow the guidelines by Hyman and Honorton, so he had three postgraduate students at Cornell University to judge to what degree the experiments followed the guidelines. The analysis and result from this comparison (the degree of similarity to the model was between 1 and 7, with 7 for maximal similarity) did show that the studies that deviated much from the model did influence the Milton and Wiseman results in a negative way (Bem et al., 2001; Palmer, 2003; Palmer & Broughton, 2000).
Applications of Electrodermal Activity (EDA) in parapsychology
Another large area of research in parapsychology concerns Direct Mental Interactions with Living Systems (DMILS) (and later the field of „remote staring‟ will be described), a method that brings us closer to applications with EDA. Beloff (1974) is accredited (Ramakers et al., 2005) as being the first to suggest the use of a physiological measure in parapsychology, and Morris (1977) carried such a study some years later. Most studies have used a conscious response measure, so EDA was a new step, being a (mostly subconscious) measure of a potentially psi-mediated response.
Although Beloff and Morris‟ work preceded this work, the standard design was established by William Braud and his colleagues during the 1970s and 1980s (Braud, 2003; Braud & Schlitz, 1991; Schlitz & Braud, 1997). This is the so-called DMILS design in which one person, the agent, systematically tries to influence a psychophysiological response of a receiver located in another room (Delanoy, 2001).
The agent and the receiver are sensory isolated from each other in order to preclude any currently understood means of sensory communication between them. The goal is for the agent to influence the psychophysiology by the receiver, as measured by electrodermal activity, EDA. The goal for the agent can be either to calm or activate the receiver‟s EDA at different times during the experimental session. Normally, a session lasts for 20 minutes, including 40 periods, each of 30 seconds duration (Delanoy, 2001).
Of the 40 periods, 10 are „activate‟ and 10 are „calm‟ periods, with the 20 „rest‟
interspersed between the activate and calm interaction periods. The receiver is held blind to the ordering of the agent‟s sending periods, so s(he) does not know when the
different periods are, but s(he) does know that each interaction period will be followed by a „rest‟ period, when no interaction with the receiver will be attempted. In some studies the agent can follow the ongoing tracing of the receiver‟s EDA giving them a possibility to see if the intention to calm down or activate do result in lower or higher activity in the EDA.
The EDA activity is measured for each period. All the data from a session is collapsed into a single session score and these session scores are used as the units of analysis. The most commonly used method is a ratio method called „percentage influence score‟, or PIS (Braud & Schlitz, 1991). This score is calculated by dividing the sum of all data of the experimental condition by the sum of all data of both experimental and control condition. In the absence of any effect, the PIS is 50% and each deviation from that mean reflects an experimental effect (Schmidt & Walach, 2000). The idea with PIS is that in the absence of a psi influence there should be no significant difference in the mean EDA amplitude between the groups.
Of the 37 DMILS studies that Braud conducted from the late 1970s to the early 1990s, 21 gave a significant result (Braud, 2003; Delanoy, 2001). While chance would give a 5 percent hit rate, the experimental finding here was in the order of 57 %, giving strong support for an analogue of mentally induced healing in a laboratory setting. Delanoy (2001) summarizes:
“Braud‟s group explored seven different living systems including two psychophysiological measures (EDA and blood pressure), various human and animal behaviour responses and the haemolysis rate of in vitro red blood cells when osmotically stressed. In assessing the overall outcome of these 37 studies, Braud and Schlitz (1991) reported a combined Stouffer z=7.72 (p=2.58 x 10-14, one-tailed). The mean study effect size (r) was .33 and 57 % of the studies were independently significant. --- The research from this lab involved 655 sessions, 449 living „receiver‟
systems, 153 different agents and 13 different experimenters” (p. 34).
Attempts to replicate this initial work by Braud using EDA as a response system have been carried out at six different laboratories (Delanoy, 2001). The replications have used either the calm/activate approach or the remote staring protocol which is to be described later. Schlitz and Braud (1997) found, in a review of this literature 19 calm/activate and 11 remote staring EDA DMILS studies. Delanoy (2001) summarizes:
“Seven, or 37 %, of the 19 calm/activate studies had achieved independently significant outcomes, with the Stouffer z = 4.82 (p= 7 x 10-7) and a mean study effect size (r) of 0.25. If all the 30 EDA DMILS studies are combined (i.e. both the calm/activate and remote staring studies), 47 % or 14 of the 30 studies obtained an independently significant outcome. The combined Stouffer z of the 30 studies is 6.17.
The associated probability is 4.58 x 10-10 and the mean study effect size (r) is at 0.25”
(p. 34).
Nevertheless despite this promising outcome, the research with EDA has been criticized, to which we will return later.
EDA and remote staring
Another application of EDA in parapsychology is the area „remote staring‟. These studies (e.g. Braud, Shafer, & Andrews, 1993a & 1993b) are very similar in design to the calm/activate DMILS studies, except that the „activate‟ period consists of the agent staring at the real-time image, shown on a monitor screen, of the receiver which is conveyed to the agent via a closed-circuit video camera system. The „calm‟ periods become „non-staring‟ periods where the agent does not view the image of the receiver.
Delanoy (2001) concludes that the replications at other laboratories of Braud‟s work also have been on the whole successful. Of the 11 remote staring studies, seven or 67 % obtained an independently significant outcome, with a Stouffer z = 3.87 (p=5 x 10-5), and a mean study effect size (r) of 0.25 (Delanoy, 2001).
In evaluating why this research seems to have been so successful, Delanoy (2001) regards the ideas put forward from Braud and Schlitz (1989 and 1991) as sound and logical, with the underlying idea that a physiological measure short-circuits the cognitive processes. Thus autonomic responses might be subjected to less cognitive interference than other types of psi responses, especially when those are based on a conscious response. Another reason is that living targets may have a greater lability than other systems, an ability to change in response to external influences. Still another reason might be that there are several potential psi sources that might contribute to any apparent psi effect: there could be telepathy between the sender and the receiver including the intention from the receiver.
The experimenter effect
Some parapsychological researchers very often get significant results, while others are almost never successful (MacFarland, 1938; Nicol & Humphrey, 1953; Pratt & Price, 1938; Smith, 2003; West & Fisk, 1953; Wiseman & Schlitz, 1997, 1999). To test the possible role of the experimenter for the result, two studies have been conducted by one sceptic and one proponent of psi, Richard Wiseman and Marilyn Schlitz respectively.
They have very different track records in obtaining evidence for psi in studies on
„remote staring‟. Wiseman has conducted four „remote staring‟ studies using EDA (Wiseman & Smith, 1994; Wiseman, Smith, Freedman, Wasserman, & Hurst, 1995), but found no evidence for a remote-staring effect (Smith, 2003). In contrast to this, Schlitz has a long record of obtaining significant outcomes in EDA DMILS studies (e.
g. Braud & Schlitz, 1991; Schlitz & LaBerge, 1997). In order to further explore the role of the experimenter in these studies, Wiseman and Schlitz jointly conducted two studies with EDA DMILS „remote staring‟ (Wiseman & Schlitz, 1997; Wiseman & Schlitz, 1999). Both studies used the same protocol. Wiseman and Schlitz each served as the experimenter, as well as the agent, in half the study sessions. They used a closed-circuit video camera system to enable the agent to stare „real-time‟ at the receiver. The first study was conducted at Wiseman‟s laboratory at University of Hertfordshire in England, the second at Schlitz„s laboratory at the Institute of Noetic Sciences in California.
Delanoy (2001) summarizes:
“In both studies, with Wiseman acting as the experimenter, the participants produced non-significant outcomes (study one: Wilcoxon z = -0.44, df=15, p=0.64, two-tailed, effect size r= .14; study two: n=35, Wilcoxon z = -0.39, p=0.69, two-tailed, effect size
=-0.7). In contrast to this, the participants which Schlitz worked with, produced
significant outcomes in both studies (study one: Wilcoxon z = -2.02, df=15, p=0.04, two-tailed, effect size r= .50; study two: n=35, Wilcoxon z = -1.93, p=0.05, two-tailed, effect size =-.33)” (p 37).
In discussing possible explanations for this difference, Wiseman and Schlitz (1997 and 1999) considered and rejected as unlikely several non-paranormal hypotheses, including
“experimental artefact, undetected sensory leakage, participant cheating, experimenter fraud and that by chance Schlitz ended-up working with more gifted psi receivers than did Wiseman” Delanoy (2001, p. 37). They mention two hypotheses that were not rejected: 1) Schlitz being more skilled in eliciting psi ability from the receiver than Wiseman; and 2) the experimenters may be the psi source in the study, with Schlitz having more psi ability than Wiseman. Smith mentions (2003) in his analysis of this experimenter effect:
“One possible explanation for this difference is that the experimenter‟s attitudes towards psi are communicated to their participants during the briefing stage of the experiment. Thus, the experimenter‟s overtly expressed attitudes towards psi may influence participants‟ motivation or expectations of success in their experiment, which in turn impacts upon participants‟ psi scores” (p 75).
This is a supposition that does have some support in the evaluation following the experimental sessions when the participants were asked to report their belief in psi:
Schlitz‟s participants reported a stronger belief than did Wiseman‟s. The published interviews with Wiseman and Schlitz also revealed how Schlitz clearly put more effort into creating a rapport with participants than did Wiseman (Watt, Wiseman, & Schlitz, 2002).
The EDA research analysed
Schmidt and Walach (2000) are very critical to the use and reporting of EDA in studies on DMILS and remote staring. Since the quality of EDA recording is crucial for these studies, they made a review of all the published reports in these fields, and compared them with the standards in psychophysiological journals, established in the 1970s and published in the beginning of the 1980s (Fowles et al., 1981; Lykken & Venables, 1971;
Venables & Christie, 1980). They compared 24 DMILS/Remote staring studies with a sample of 39 recent psychophysiological studies published in Psychophysiology and International Journal of Psychophysiology. They came to the conclusion from this, that parapsychologists do not meet the current standards. According to Schmidt and Walach (2000) there is not even one study conducted by parapsychologists which refers to psychophysiology‟s measurement standards published in 1981. They maintain that theses studies may either contain artefacts, or, on the other hand, may not detect the supposed effects. They mean that there have not been any efforts to understand the results of EDA experiments or to address the origin of the irregularities in detail. Most of the reports do not use the required SC-technique (constant voltage method) or the appropriate electrode paste. The scoring method remains unclear in more than half of the studies and when additional knowledge from the overviews or meta-analyses is not available, it is not possible to interpret the reported effects.
For those involved in parapsychological research, it is a well-spread opinion that even if the results are highly debated and controversial, there is almost a consensus that the
methods used in parapsychology during the last decades have high standards. This well- spread opinion can of course be questioned when you are confronted with the critical comments from Schmidt and Walach (2000). Does this mean that the standards after all not are that high, that there are reasons to question not only EDA-applications, but also other parts of the field? Or are there reasons why the experiments that Schmidt and Walach have studied are an exception?
Walach comments this (personal communication, April 22, 2008): “the people who originally developed the EDA measurement protocols were quite knowledgeable (Schlitz, Braud) and most of the ones who are now using EDA (Broughton, Watts) probably use it the right way. There was a large series of experiments done in Edinburgh in the late 80ies beginning of the 90ies, where the proper knowledge seems to have been lost, and I am not confident that they were carried out rightly. I am quite sure that our paper stirred whoever was responsible then into correcting those methods.
Otherwise, I think parapsychologists are far more meticulous than standard scientists. It is often that when they use methods that need some specialised knowledge that this knowledge is not always known to all. But now this seems to be well known.”
Schmidt adds (personal communication, April 28, 2008): “Maybe I can add that in my view the paper had quite an effect on the psi community. E.g. now almost everybody uses the right nomenclature, they speak of EDA and skin conductance instead of galvanic skin responses or the like. Also authors spend more time to explain their methodology and are aware that this is a sensitive issue. Regarding the analysis methods this is still a somewhat mixed situation where researcher sometimes use standard psychophysiological methods but also very often their own parameters created from the data.”.
Deborah Delanoy, Northampton University and a leading researcher in the field adds (personal communication, April 21, 2008): “Parapsychology work is looking at very different things from straight psychophysiological work, so criticisms of one do not necessarily apply to the other. Some of Stefan et als., comments ignored the fact that if there was any undetected problem in the measurement, it would have impacted equally upon all the condition of the study. As the psi outcomes are based on „between condition‟ comparisons (an important point), something that would affect all the conditions equally should not have an impact on the psi outcome. On the other hand, ensuring adequate randomisation is crucial as there are cross session changes (due to habituation, etc.) that can impact upon early session conditions differentially from later session conditions.“
Caroline Watt, Edinburgh University and also a leading researcher in the field adds (personal communication, April 21, 2008): “I would note that Stefan's (Schmidt) critique in many cases does not suggest the introduction of specific artefacts that would produce an illusory appearance of psi. Rather, failure to follow the highest standards in EDA methodology is likely to introduce noise and random artefacts that might obscure any psi signal.”
So it seems it is not obvious that the research that Schmidt & Walach criticize can easily be disregarded. Comments from both Watt and Delanoy indicate this is the case. This research is still to be debated.
Details that are considered important by Schmidt and Walach (2000) are the kind and size of electrodes, the kind of electrode paste applied, the voltage used, the site of the electrodes, the time lag between electrode placement and start of data recording, the site pre-treatment, the climate conditions, how the different possible artefacts like breathing, movement of hands were dealt with and if the tonic or the phasic component of EDA was used. Schmidt and Walach (2000) found that seven of eight studies published since 1997 have used only the tonic component, a shift which was never discussed. They also comment that the PIS score (Percent Influence Score) was often used, but since it is standardized by the mean (50%), but not by the standard deviation, different PIS values cannot be compared. A PIS of 51 % with a small SD might reflect a larger effect than a PIS of 60% with a large SD.
There are different opinions when it comes to the question what variable in EDA is most suitable for detecting remote effects. The most widely used parameter is the mean level of the SC, but in a more recent report, Schmidt, Schneider, Binder, Bürkle and Walach (2001) argue that this is not the most appropriate variable to detect psi and distant effects. Because the recording, processing, parameterization, and evaluation of the EDA data are the crucial point in interpreting them, they studied the EDA as a function of different methodological approaches. They compared different EDA parameters (tonic and the fast-changing phasic) and found the effects to be similar in size. This indicates that the effect is more of a global one rather than a very specific influence. They conclude that the activation intention effect is related to a more global physiological state reflecting overall arousal (affecting several physiological systems).
Their results also indicated that a psychophysiological way of parameterization of the phasic EDA data was more appropriate. Their data clearly revealed, they argue (Schmidt et al., 2001), that their method of using the classical psychophysiological parameters outperformed the method traditionally applied in DMILS/remote staring research. The classical parameters are the numbers of NS.SCRs and the sum of amplitudes of NS.SCRs, SCR standing for skin conductance response, the reaction in the skin on a stimuli when the conductance for a short while increase and then goes back to normal. NS stands for non-specific or spontaneous, NS.SCR standing for non- specific skin conductance reaction (Boucsein, 1992), phasic parts of the EDA which cannot be traced to any specific stimulation, while otherwise the term “reaction” for phasic electrodermal phenomena suggests that there is a distinct relationship to a stimulus producing an EDR.
If this is confirmed in further replications, then it would appear that all the experiments using the phasic EDA have underestimated the effect because of the inappropriate parameterization. In this pilot study, they also wanted to study if there were differences between different statistical procedures used in DMILS/Remote staring literature. Could any technique be regarded as being most appropriate? They found that it is obviously inappropriate to use the PIS method:
“This conclusion applies to both the interpretation of PIS as a score of amount of influence expressed in a percentage value as well as to the significant testing based on these scores. We highly recommend abandoning this method for psi research”, they conclude (Schmidt et al., 2001, p 79).
Moreover according to the authors, the other method, the two-component model (combining two statistical tests, Wilcoxon signed ranks test for pairs of data, followed
by a test of significance of the whole experiment is applied), is not an appropriate procedure, because a basic assumption of the test is violated (the test implies that pairs to be compared are dependent, but it assumes that the pairs generated as the basis of the comparison are themselves single, independent measurements. With all pairs of data to be tested stemming from the same participant this assumption of independence is violated). Instead, the authors favour the Wilcoxon signed rank test as the simplest method, since it is easy to calculate and free of any distribution assumptions. This might be important, they argue, because EDA data from different participants cannot be guaranteed to have homogenous variances. If the difference scores between activate and calm epochs are normally distributed, the assumptions for a paired t test are fulfilled.
This parametric testing procedure might result in a slight increase in statistical power.
As will be seen later, in this study, Wilcoxon signed rank test was planned, but since the variance was rather normally distributed, a t-test, paired samples was used.
In a re-evaluation of two SC studies (Delanoy, Morris, Roe, & Brady, 1999; Watt, Ravenscroft, & McDermott, 1999), Stevens (2000) found indications that the variance of the SC might be a more useful parameter then the mean level. He found that the skin- conductance responses recorded during influence periods were significantly more variable then during the rest period in both studies, the calm period showing the highest variability, then the activate period, then a large drop for the rest periods. Based on the variance of electrodermal activity, there were significant differences between any type of influence attempt and rest periods (p<0.01 and p<0.002, both 2-tailed for the two DMILS datasets used).
Following the critique by Schmidt and Walach (2000), there will inevitably be an uncertainty as to what weight should be attributed to the findings of the studies in the fields of DMILS and remote staring. We now turn to other methodologies used to detect emotions.
Other techniques to detect emotions
There have been some studies in parapsychology aimed at detecting the presence of emotions using physiological measures other than the EDA. Radin and Schlitz (2005) used cutaneous EGG (electrogastrography), a non-invasive way to monitor the gut‟s myoelectrical behaviour, to detect emotions. EGG frequencies and amplitudes closely correlate with stomach contractile measured with invasive electrodes. They wanted to study if gut feelings, commonly reported visceral sensations that are virtually synonymous with intuitive hunches, may involve information gained by non-ordinary means. Because of the close relationship between gut feelings and emotions, they especially tested whether or not a person‟s gut feelings might respond to the emotions of a distant person. Their result appears to support the hypothesis that one person‟s gut feelings can respond to a distant person‟s positive and sad emotions. They discuss some alternative explanations (chance, inappropriate statistics, sensory cues, expectation bias, and physiological drift), but regard them all as implausible.
Arousal, physiological detection of emotions and EDA
During the 1880s, psychological factors were first being observed in relation to electrodermal phenomena (Boucsein, 1992). Since then this phenomenon has been one of the most used biosignals in psychophysiology, even if not all details in the
electrodermal phenomena are completely understood. The term electrical activity in the skin, now called EDA (electrodermal activity, formerly called GSR, galvanic skin response) was introduced 1966 by Johnson and Lubin (as mentioned by Boucsein, 1992) as a general term for all the electrical phenomena in the skin, both active and passive.
“The essential discovery of these phenomena is attributed to two researchers who might not have been aware of each other, the French neurologist Féré (1888), and the Russian physiologist Tarchanoff (1889). Féré used an external direct current, and observed a decrease in SR (skin resistance) following sensory or emotional stimulation in hysterical patients” (Boucsein, 1992, p 4).
Tarchanoff did use the endosomatic methods (without an external current applied) and observed electrodermal changes not only following sensory stimulation and voluntary muscle contractions, but also imagination, mental arithmetic and expectation.
Tarchanoff believed the electric activity to be a result of sweat gland activity. It was however Richter who 1929 “was the first to state the hypothesis of a causal mechanism for EDA, including both epidermal and sweat gland mechanisms, which is still regarded as valid” (Boucsein, 1992, p.5; also Edelberg, 1972a; Fowles, 1986). Today there is also theoretical as well as empirical evidence for close connections between higher stages of information processing and certain EDA parameters (e. g. Bagshaw, Kimble, &
Pribram, 1965; Boucsein, 1992; Edelberg, 1972b; Pribram & McGuinness, 1975 and 1976) and there may even be relationships of EDA with storing and retrieval of memories (e. g. Boucsein, 1992; Raskin, 1973; Öhman, 1979).
EDA is regarded (Boucsein, 1992, p. 263) “as a sensitive and valid indicator for the lower arousal range, reflecting small, mostly cognitively conditioned variations in arousal”. Boucsein (1992) asserts that EDA measures are highly applicable to emotions and stress research. He bases this on the EDA being mediated solely by the sympathetic branch of the ANS making it not subject to peripheral parasympathetic influences as most other autonomic variables are:
“Consideration should be given to the use of different EDA parameters as markers for the various emotional states, as for example in research on emotional expression where EDR amp. (the amplitude of EDR, the response or reaction in EDA) is correlated with the inner emotional involvement while HR (heart rate) is more likely to reflect overt emotions”, Boucsein argues (1992, p. 373).
We can here observe that there are different opinions as to what parameter is most appropriate to use to detect inner emotions in the use of EDA in parapsychology – Schmidt et al. (2001) suggest variance to be the best parameter, Boucsein (1992) mentions that the amplitude is correlated with the inner emotional involvement, but does this mean that the mean level is the best parameter? This is not clear. Could it be that both are right, that both parameters are possible, but variance is better? Is it possible that Schmidt et al has done more detailed research then Boucsein on this special issue and therefore has drawn another conclusion? Or does Schmidt‟s research have a background in ordinary psychophysiology and with difficulties in applying them to parapsychology, as does Delanoy indicate (personal communication mentioned above)?
This difference remains to be analysed and understood. The issue is probably still not finally solved.
Cacioppo, Berntson, Klein and Poehlmann (1997) conducted a meta-analysis providing data relevant to the question whether or not emotion-specific autonomic patterning exists. They showed that
“skin conductance level increased less in happiness than in disgust, but disgust did not differ from control conditions. --- Fear was associated with greater increases in non- specific skin conductance responses (i. e. spontaneous reactions – or as mentioned before, NS.SCR = non-specific skin conductance reaction) and smaller increases in skin conductance level than sadness” (Cacioppo, Berntson, Larsen, Poehlmann, & Ito (2000, p. 183).
The authors also mention that a variety of theoretical and empirical work suggests that, all else being equal, negative emotions may be characterized by greater autonomic activation than that found in positive emotions (Cacioppo et al., 1997; Taylor, 1991).
The meta-analyses indicated
“that even a limited set of discrete emotions such as happy, sad, fear, anger, and disgust cannot be fully differentiated by visceral activity alone, but follow-up meta- analyses in this literature did suggest that the negative emotions are associated with stronger ANS responses than are the positive emotions” (Cacioppo et al., 2000, p.
184)
a fact that indeed does support the use of negative pictures, as is the case in this study.
Thus, the evidence for the visceral differentiation of emotion, like that for incipient facial differentiation, is clearer when positive and negative emotions are contrasted than when discrete emotions are contrasted.
EDA is today used in many applications to test the reaction in an organism to external stimuli, for orienting and habituation to events that possibly could be a threat. This orienting response is a psychophysical reflex and constitutes changes in a number of physiological functions when the subject is exposed to strong emotional stimuli (Radin, 1997). The purpose of the reflex is to sharpen our perception and prepare us for the possible danger. In humans, these reactions can be elicited by emotionally provocative pictures, films etc.
Factors influencing reactions on looking at pictures
Lang, Greenwald, Bradley and Hamm (1993) reported a study where they wanted to examine the organization of emotional perception as instigated by a broad range of static visual stimuli. The question concerned if the emotional response is organized along two strategic dimensions: those of affective valence and of arousal. Coloured photographic pictures that varied widely across the affective dimensions of valence (pleasant-unpleasant) and arousal (excited-calm) were each viewed for a 6-s period while various measures such as skin conductance were being carried out. Judgements relating to pleasure, arousal, interest, and emotional state were recorded. The study found there was a significant covariation between skin conductance magnitude and arousal ratings. Interest ratings and viewing time were also found to be associated with arousal. Furthermore, it was found, as predicted, that skin conductance response increased monotonically with ranked arousal. Seventy-seven percent of subjects showed
positive conductance/arousal correlations and for 33%, the relationship was statistically significant. In an analysis including all the measures, two factors were clearly identified: (a) an Affective Valence factor based on facial muscle responses, heart rate, and pleasantness judgements, and (b) an Arousal factor, consisting of skin conductance response, viewing time, and judgements of interest and arousal.
Greenwald, Cook III and Lang (1989) examined the relationship of verbal affective judgements to psychophysiological response organization, based on a dimensional analysis of emotion. Forty-eight unselected undergraduates viewed 21 coloured photographic slides, which had previously been shown to vary widely across the affective dimensions of valence and arousal. During the viewing interval, measures of skin conductance, facial electromyographic and heart rate measures were taken. The study showed, by regression analysis, that larger mean skin conductance changes were significantly related to increased arousal ratings. This arousal/skin conductance relationship in the overall sample showed highly significant linear and quadratic trends.
EDA and remote emotion
Case reports in parapsychology indicate that emotions are important in apparently paranormal experiences (Delanoy, 1989), but their role is not very well understood.
Most studies have used a conscious response measure, and most studies using EDA have been carried out in order to study remote influence (like in the DMILS protocol).
In fact, there are very few studies that have used this measure for investigating whether or not it is possible to detect the effect of emotions, originating from another distantly located person. Since there is a close association between conscious emotional experience and physiological arousal, it makes sense to study remote emotion using physiological responses. Using this measure enables us to avoid cognitive bias (Cacioppo, Berntson, Larsen, & Ito, 2004) since the inclusion of additional (potential) cognitive correlates may inhibit the psi process we are studying.
Ramakers (Ramakers et al., 2005) mention two studies of remote emotion using physiological measures. One of them, using skin conductance, found higher activation for positive pictures (i. e. positive emotions) compared to the neutral condition (Delanoy
& Sah, 1994). This study is also one of the few (also Targ & Puthoff, 1974; Tart, 1963) that was designed to explore whether conscious responses or unconscious responses (EDA) would elicit a larger ESP effect. The second study is one already mentioned, the study by Radin and Schlitz (2005), that used another physiological measure.
Another question of interest in these experiments (as well as for all telepathy experiments using pictures as target material) is whether to use positive or negative pictures. Delanoy and Sah (1994) refer to an unpublished, undergraduate psychology project (Boswell, 1993) supervised by Delanoy that further explored the use of emotional targets. The study found significant overall psi-hitting, significant scoring for the sending of positive emotions and nearly significant scoring for neutral emotions, with negative emotions eliciting scoring very near MCE. The results from this study suggest that positive emotions may be better as targets than negative emotions, but are not vastly superior to neutral ones. On the other hand, in a dream clairvoyance study by Dalton, Steinkamp and Sherwood (1999), it was found that emotional targets, particularly when they were negative, were better targets (Sherwood & Roe, 2003). This is perhaps not surprising given that spontaneous cases of dream ESP often seem to
feature negative life events (e. g. Ullman, Krippner with Vaughan, 1989) and that emotions seem important in spontaneous psi experiences from the work of L. E. Rhine (1967) and her collection of subjective experiences.
The present study examines the effect of remote emotion on the skin conductance of the receiver, when a sender is viewing emotional pictures. It is a conceptual replication of an experiment by Ramakers et al. (2005). His study is shortly described above. The result of his study showed:
“no significant difference between the emotional and the neutral conditions and no effect of the combined emotional conditions on the mean variance. A differential analysis of the positive, negative and neutral conditions showed that mean log skin conductance was highest in the negative condition, but not to a significant degree”
(Ramakers, 2008, p. 21),
as he summarizes it in the final report. He also notices “a strong trend for the mean variance to be lower in the positive condition (λ2 = 8.197, df=2, p=0.017, two-tailed)”
(Ramakers, 2008, p. 21), and that “the results revealed a pattern of higher activation in the negative condition” (Ramakers, 2008, p. 21). Of interest here is that mean log skin conductance was highest in the negative condition, but not to a significant degree, and a pattern of higher activation in the negative condition, two important aspects in his study.
There are two important differences between their study and this present one. The first difference was that Ramakers and his co-workers used 5 positive and 5 negative pictures (and 10 neutral pictures), while the present study involved 10 negative pictures and 10 presentations of a blank screen. The reason for this was that using only one kind of emotional picture (instead of splitting up in two kinds) would give more statistical power (with more trials it would be easier to get significant results). It was not easy to decide whether to choose negative or positive pictures for this study, since there is support for both kinds as being successful target as psi stimuli. The choice was made for the use of negative pictures, because many researchers have reported some success with this choice (Bem, 2003; Dalton et al., 1999; Radin & Schlitz, 2005). It might be that success depends on in what type of experiment the pictures are used, as some have employed emotional targets for PK studies and others for ESP (McCraty, Atkinson, &
Bradley, 2004; Radin, 2004 and the studies reviewed here). There may also be a difference between telepathy and precognition with negative targets favouring precognitive designs. Given that if one can anticipate or “precognize” danger (pictures with threat or disgust) then one would have a better chance at survival (Bem, 2003;
Radin, 1997). Support for this choice also comes from what is previously mentioned (Cacioppo et al., 2000), follow-up meta-analyses in this literature that did suggest that the negative emotions are associated with stronger ANS responses than are the positive emotions. Beside this, it is also worth recalling the research mentioned above (Cacioppo et al., 1997) suggesting that, all else being equal, negative emotions may be characterized by greater autonomic activation than that found in positive emotions, and thus being better for EDA research.
The second difference was that Ramakers measured EDA for both the receiver and the sender, while in the present study (mainly because of problems involved in arranging two sets of equipments), EDA was only measured for the receiver. The EDA- measurement for the sender was afterwards partly supplemented by a post-session
questionnaire asking the sender to estimate how much they reacted to the pictures, thus making it possible to compare this estimation with the EDA-measure for the receiver.
The hypothesis
The singular prediction here was that emotional target pictures viewed by a sender will alter the standard deviation of the receiver‟s SC. Influenced by Stevens (2000) and Schmidt et al. (2001) the choice was to use the standard deviation of the SC as the dependent parameter instead of the mean level of the SC. The hypothesis was that the mean standardized standard deviation of the SC in the periods when an emotional picture is shown would be different from the mean standardized standard deviation for rest periods, when no picture is shown.
Method Participants
Thirty-one pairs participated (59 individuals) in 60 experimental sessions, fulfilling the pre-set target number of trials. It was decided that the person who was first in the sender role would change and carry out a second session in the role of the receiver (and vice versa). There were 17 male and 43 female receivers. The mean age of all receivers was 39 years (SD=15.7 years, range 14-70 years). Most participants were recruited to the experiments following an advertisement in Dagens Nyheter, a large circulation Swedish daily newspaper, giving 44 persons (23 pairs, 45 sessions). The advertisement sought foremost persons, who could bring a partner or friend, with whom they had an emotional or biological relation. Some participants were also recruited from members from the Swedish Society for Parapsychological Research (giving 9 persons, giving 5 pairs, 9 sessions), as well as persons reading NewAge websites (giving no person).
Finally, a call for participants was also placed on the notice board at the Department for Psychology at the University, giving 6 persons (3 couples, 6 sessions), and a few experiments used as pilot experiments to test the procedure.
The relationships between the pairs were as follows: 16 friends, 15 parent-child, 2 brother-sister, 1 other family-relation, 22 married/cohabitants and 4 others. Eighteen pairs had a biological relationship.
On the occasions when pairs of participants contributed two sessions (there were altogether 26 pairs contributing 2 sessions each, two pairs parent-child contributing one session each – the children being too young, and one family – parents and two children – contributing 6 sessions), it was decided that the person who initially functioned in the role of the sender would then change for the second session and take then the role of receiver (and vice versa). The operant rule was that if both persons in a pair were over 18, they carried out two sessions, otherwise only one. For ethical reasons those under the age of 18 were not allowed to view the pictures.
Material and Equipment
The experiment took place in the basement of the Department of Psychology, Stockholm University, see figure 1 for laboratory layout. The two rooms were old air- raid shelters, with two heavy, thick (together 35 cm) iron-doors between them, thus providing a good means of sound reduction. The EDA equipment was placed in the
inner room, while the sender was positioned in the outer room, in front of a computer that would show target pictures. Relaxation music by Björn J:son Lindh was played over the headphones of the receiver, which would additionally serve as a means of minimizing disturbing sounds from the external world, such as the at times noisy corridor above. There was approximately 4 metres space between the sender and the receiver. The distance from the wall to the table for the receiver was 1.85 metres.
SC from the receiver was recorded on two 24 bit serial port model devices from Biopac Systems Inc., model mp 100A, linked to a Power MacIntosh G3 computer by means of computer software AcqKnowledge III for MP100WS (version 3.2.). Electrodes of the EL204S using 8 mm Ag/AgCl type were used with isotonic paste as electrolyte (0.5%
NaCl/100ml H2O).
The 10 negative pictures were selected from the International Affect Picture Systems (IAPS) and mixed with 10 blank exposures following a suggestion from Ramakers (personal communication, 2006-01-23), who had noticed that their neutral pictures might not be functioning as neutral stimuli. The 20 “pictures” were arranged in blocks of 10 with 2 in each. Every block contained one negative and one blank picture. The order within each block concerning which one that would be negative, was randomised, a choice carried out by the computer. In this way, expectancy effects would be avoided, since the receiver would then not know when a negative picture was shown. The programming for showing the pictures was done by the supervisor Joakim Westerlund, JW. The same pictures were used for each session. The choice of pictures to be included in the study was done by the author and JW. Following Ramakers (Ramakers et al., 2005), we selected to use pictures from the IAPS that were rated high on arousal and low on valence. These values were for all IAPS-pictures found in a normative table for all subjects (there were also tables for male, female and children), provided by Lang, Bradley and Cuthbert (1999). The aim here was to select pictures that would not evoke a high level of disgust. This would ensure that the agent during the rest period would be able to return to a relaxed state and then, with the next picture, is prepared to tune in to next emotion in accordance with the varying content of the pictures.
In the study, a questionnaire was used, that was similar in content to that of Ramakers, with one pre-session section and one post-session section; The changes concerned the post-session questionnaire in which the subjects were asked to give their reaction to the pictures shown and a pre-session section of the questionnaire which assessed belief in telepathy, subjective paranormal experiences and asked for an assessment of the type and intensity of the relationship between sender and receiver. A post-session section asked the sender to rate their (emotional) reaction to each of the different pictures. The receiver‟s response was measured purely physiologically. The experimenter would later have the possibility of matching the EDA of the receiver with the ranks of the sender and also with value of arousal and valence of the pictures.
The author was the experimenter for all sessions. A description of the design of the study had been given earlier at the first contact by telephone or email. Participants were informed that some pictures were very negative and that they were free to withdraw from the study at any time. When arriving to the laboratory, both participants were shown the sender and the receiver rooms. They were told that there were 10 blocks of pictures with two pictures in each, one emotional and one neutral (designated as emo- neu or neu-emo), the choice being made by the computer. Participants were then given
Figure 1. Laboratory layout.
Procedure
questionnaires and asked to fill out the pre-session questionnaire. The participants decided who was to take the role as sender in the first session. Since in many cases the decision seemed to be a fairly arbitrary one, both questionnaires were laid on a table, one marked for who would first be the sender, the other for the receiver. If participants had not decided on their initial role in the study when they sat down, they chose the questionnaire in front of them, and in so doing selected their role. The sender was asked to pay full attention to the pictures and try to immerse themselves into the emotionality of each picture. They were asked to try to maintain the emotion for as long as the picture was on screen. The receiver was asked to relax and to try to keep his or her mind blank and open to the possibility of getting any impression. The receiver was placed at the table in the inner room, the fingers were cleaned, gel was put on the electrodes and these were attached on the non-dominant hand (on the index and middle finger) with a Velcro band. Following this, the person was asked to sit with uncrossed legs and be as calm as possible. Headphones were placed over the receiver‟s ears, the relaxation music by Björn Json Lindh was started (and played during all the session) and the volume adjusted. The experimenter then started the EDA equipment and checked for the occurrence of a response from the receiver by asking the receiver to take a deep breath or to cough. It didn‟t seem likely that the external noise (due to sounds from the corridor above) was not noticeable when music was played through the headphones.
The sender sat down in front of the computer screen in the other room, so as to sit comfortably and to be able to relax. Noise reduction earphones were placed over the sender‟s ears. After starting the recording, the experimenter marked the time of the start
Receiver Mac
computer
EDA
Sender Screen
Sofa for experi- menter Table for
questio- nnaires
Door to come in Compu- ter
on a special channel on the Mac. He then left the room and closed both iron-doors between the rooms. The computer-programme was started 30 seconds after the time mark was made, thus enabling the synchronisation of the EDA equipment with the computer timing. This preparation gave the receiver just over 1 minute to relax (while in the Ramakers design they had about 5 minutes).
Finally, the experimenter sat down in a sofa diagonally behind the sender and behind a film screen (so as not to disturb the sender). Each picture slide including the blank was presented for 30 seconds and followed by a recovery period of 15 seconds. Following the completion of the first session each pair of participants exchanged roles, and then the second session started. After both sessions, they were debriefed, filled in a post- session questionnaire. Participants not being psychology students also at the end received a token payment in the form of two tickets for a movie. For psychology students this participation was part of their training.
Security Precautions
As mentioned before, the 20 “pictures” were in 10 blocks, containing 2 in each. Each block contained one negative and one was neutral, i. e. blank slide. Each block could be designated Neg-Neu or Neu-Neg. The first negative picture slide was always positioned in the first block, as the first or second picture, etc. The selection as to which picture in a block that would be the target picture and which would be a blank was determined by a computer programme for randomisation (based on the random number generator, Visual Basic pseudorandom algorithm). The outcome of the selection was saved in a text file in the computer. As a further security measure, on completion of the session, this file was immediately emailed to the supervisor JW. The experimenter was held blind as to the periods on the EDA-curve that corresponded to an emotional picture and what was a blank picture.
Analysis
Following the completion of all 60 experimental sessions, the mean value of SC as well as standard deviation (SD) was calculated (partly automatically, partly manually – by the author marking on the curve for each period of the 30 s and have the Mac to calculate these two measures) for each real picture as well as blank picture. These data were also sent by email (as an excel file) to JW, who after all experiments were completed decoded the file containing information about the order the pictures were shown. For each session, the SD of the SC was calculated for each of the 20 pictures (10 emotional pictures and 10 blanks). These 20 SDs were then standardized by transforming them into z-scores using the following formula:
SD
SD SD
SD z SD
Results An alpha of 5 % was used in all analyses.
The hypothesis
The mean standardized SD in the skin conductance for the emotional pictures was .00599, while the mean standardized SD for the rest periods was -.00599. (The SD of the SDs was .1757 for both emotional pictures and for rest periods).
